{"id":"906a7e6c-c959-451b-91fe-8205958baf61","arxiv_id":"2505.01714","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"In Rayleigh-Bénard convection, Lagrangian pair statistics show kinetic energy cascading downscale at small scales and upscale at large scales, with a gradual mixed regime in between.","lead":"Using 3D particle tracking in a water-filled Rayleigh-Bénard cell, the authors show that kinetic energy flows from larger to smaller scales at small separations and reverses direction at separations near the Bolgiano scale. The work provides experimental Lagrangian evidence for a split kinetic energy cascade in convective turbulence, with different flow structures associated with each direction.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The sign flip in ⟨˙k⟩_r at r† may be an artifact of the unsubtracted large-scale circulation; the Lagrangian flux identification assumes homogeneity and no mean shear, so the split-cascade conclusion is unverified.","rationale":"The reader focused on the numerical estimation of ˙k from discrete trajectories, which is a valid practical concern. I find a more fundamental issue: the theoretical identification of ⟨˙k⟩_r as the kinetic energy flux is not justified for this inhomogeneous, sheared, wall-bounded flow. The paper computes all quantities from the total velocity field, including the strong LSC mean flow, and never subtracts the mean. At large separations, the relative velocity between two fluid parcels is dominated by the LSC shear, and the measured ⟨˙k⟩_r then reflects the rate at which the mean shear does work on the relative motion—a production term, not a cascade flux. The paper's own statement that the structure-function slope exceeds BO59 due to LSC shear confirms that mean shear is non-negligible at these scales. Additionally, the derivation of Eq. (4) invokes ∇⟨T⟩=0; this is valid for a closed domain but the actual measurement volume excludes near-wall regions, so boundary flux terms may contaminate the balance. This concern is directly testable: subtracting the mean flow from the particle velocities and recomputing ⟨˙k⟩_r, or comparing with an Eulerian flux estimate, would determine whether the sign flip is physical or a mean-shear artifact. Until then, the central split-cascade claim is not verifiable. I therefore adjust the verdict from CONDITIONAL to UNVERDICTED, since additional analysis, not merely a methods clarification, is required.","tokens_in":9029,"tokens_out":12361,"duration_ms":136592,"concrete_test":"Recompute ⟨˙k⟩_r from the same PTV data after subtracting the time-averaged (or low-pass-filtered) Eulerian velocity field—or removing the leading LSC mode—from each particle velocity before forming δv and k. If the positive ⟨˙k⟩_r for r>r† disappears or changes sign, the upscale cascade claim is not supported. A complementary check is to compute the Eulerian scale-by-scale kinetic energy flux (e.g., Togni et al. 2015) from the same velocity fields and compare the sign and zero crossing with the Lagrangian result.","verdict_should_be":"UNVERDICTED","load_bearing_attack":"The central claim—that the Lagrangian flux ⟨˙k⟩_r flips from downscale to upscale at r†≈0.32L—rests on identifying ⟨˙k⟩_r with the interscale kinetic energy flux. This identification (Eqs. 3–5, following Mann et al.) assumes statistical homogeneity and the absence of a mean-flow contribution. The experiment is a wall-bounded Rayleigh-Bénard cell with a strong, persistent large-scale circulation (LSC), and all statistics are computed from the total measured velocity (e.g., Eq. 6 uses δv without subtracting the local mean). At separations 0.3–0.5L, δv is dominated by the LSC shear; the authors themselves attribute the super-BO59 structure-function slope (ζ=1.3±0.07) to \"shear effects associated with the LSC.\" The positive mean ⟨˙k⟩_r for r>r† could therefore be produced by mean-shear production of relative kinetic energy rather than by an upscale turbulent cascade. Moreover, the derivation of Eq. (4) uses ∇⟨T⟩=0, which holds for a closed domain; the measurement volume excludes near-wall regions, introducing boundary terms that may bias the balance. The preprint does not discuss mean subtraction or validate the Lagrangian flux against an Eulerian budget. Without such a check, the sign flip may be an artifact of the LSC, and the split-cascade conclusion is not supported.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript reports an experimental study of the Lagrangian kinetic energy transfer in a cubic Rayleigh-Bénard cell at Ra = 2.5×10^9 using three-dimensional particle tracking velocimetry. The authors compute the rate of change of the relative kinetic energy of particle pairs, kdot, as a function of separation r and observe a negative mean at small scales, a crossover to a positive mean at r† ≈ 0.32L, and a broad transition region between the two. They interpret the sign change as a split cascade—downscale at small scales and upscale at large scales—characterize the non-Gaussian statistics of kdot, and use joint statistics with the separation velocity to argue that downscale and upscale transfer events have different flow topology. The central claim is that the Lagrangian kinetic energy flux reverses direction at a scale near the Bolgiano scale.","tokens_in":9323,"tokens_out":5654,"duration_ms":53193,"significance":"If substantiated, the result would be a valuable experimental confirmation of a split cascade in Rayleigh-Bénard convection, with implications for the scale-by-scale energy budget and for Lagrangian turbulence theory. The paper has notable strengths: it analyzes a large dataset (2.5×10^5 trajectories, millions of pairs per bin), the central observable is measured rather than inferred from a fitted model, the structure-function scalings are obtained by independent fits, and the topological analysis of the joint statistics is original. The main risk is that the flux identification and the sign of the mean kdot may be corrupted by the strong large-scale circulation, which is not subtracted and which the authors themselves identify as causing shear effects at the same scales.","major_comments":[{"comment":"The load-bearing claim that ⟨kdot⟩_r crosses zero at r† = 0.32L requires that the values of ⟨kdot⟩_r be statistically reliable and that the estimator be fully specified, but neither is provided. The manuscript does not state how the Lagrangian derivative kdot is obtained from the discrete PTV trajectories: no differentiation scheme, filter width, spline order, or time window is given, even though kdot involves particle acceleration and is therefore strongly affected by tracking noise. Fig. 2 shows no error bars or confidence intervals on the mean or the moments, so the reported r† uncertainty (0.32 ± 0.01L) has no visible basis. I request a Methods paragraph describing the kdot estimator, and bootstrap or subsampling error bars on the curves in Fig. 2, with particular attention to the zero crossing.","section":"Flux statistics (Fig. 2)"},{"comment":"The identification of ⟨kdot⟩_r with an interscale energy flux, following Mann et al., assumes a homogeneous flow with no mean shear. The experiment is wall-bounded Rayleigh-Bénard convection with a persistent large-scale circulation, and the manuscript itself attributes the super-BO59 structure-function slope ζ = 1.3 ± 0.07 (Fig. 1f) to \"shear effects associated with the LSC\" at separations 0.3L ≲ r ≲ 0.4L, exactly the range where ⟨kdot⟩_r becomes positive. Since Eq. (6) uses the total velocity increment without subtracting the local mean or the LSC, the positive mean ⟨kdot⟩_r for r > r† could be produced by mean-shear production of relative kinetic energy rather than by an upscale turbulent cascade. I ask the authors to test this by recomputing the flux after removing the large-scale/LSC contribution (e.g., subtracting a locally averaged velocity at scale r or conditioning on the LSC phase) or by comparing the Lagrangian flux with an independent Eulerian scale-by-scale budget.","section":"A Lagrangian view energy flux, Eqs. (3)–(5); Methods; Fig. 1(f)"},{"comment":"The derivation of Eq. (5) relies on ∇⟨T⟩ = 0 via Stokes theorem and on the global cancellation ϵb = ϵk, both of which hold for the full closed domain. However, all statistics are computed in a measurement volume that excludes near-wall regions, and trajectory-pair sampling is not uniform in space, so the boundary term and the work-dissipation balance inside the sub-volume are generally nonzero. Without quantifying these contributions, the step from Eq. (4) to Eq. (5) is not justified for the measured data, and the interpretation of ⟨kdot⟩_r as a pure transfer term is incomplete. The authors should either show that the omitted terms are negligible in the measurement volume or formulate the flux balance with explicit boundary terms.","section":"A Lagrangian view energy flux, Eq. (4)"},{"comment":"The topological conclusion that downscale events at large scales are associated with separating trajectories while upscale events are associated with converging trajectories is supported only by a visual inspection of the conditional means in Fig. 3. No confidence intervals or significance tests are given for the zero crossings or the inflection point of ⟨kdot|δvr⟩_r, and the number of independent samples per bin is not reported. Since this is a central secondary claim, I ask for a quantitative statistical characterization (e.g., bootstrap intervals on the conditional mean and a test of the monotonicity/inflection).","section":"Flow structure and energy flux (Fig. 3)"}],"minor_comments":[{"comment":"The phrase \"how kinetic energy is transfers across the scales\" is ungrammatical, and \"downwscale\" should be \"downscale\".","section":"Abstract"},{"comment":"The heading \"A Lagrangian view energy flux\" should be \"A Lagrangian view of the energy flux\".","section":"Section heading"},{"comment":"The notation T ≡ p v − 2ν v S mixes a vector and a tensor product; please define the contraction explicitly (e.g., v·S) and state the vector character of T.","section":"Eq. (2)"},{"comment":"The flatness values cited in the text (\"from approximately 200 at r = 0.05L to approximately 30 at r = 0.5L (not shown)\") are not displayed; either add a panel or remove the quantitative claim.","section":"Flux statistics"},{"comment":"The inset histogram lacks axis labels and a legend for the Gaussian comparison; please clarify the normalization and the parameters of the Gaussian.","section":"Fig. 2(a) inset"},{"comment":"The scaling exponents ζ = 0.67 ± 0.06 and ζ = 1.3 ± 0.07 are stated without specifying the fitting range or the fit method; a brief description would help reproducibility.","section":"Fig. 1(f)"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within scope and the split-cascade claim is timely. The main risk is LSC contamination of the flux sign; the paper would be considerably stronger if the authors add the requested LSC-removal check and an Eulerian budget comparison. I do not see a citation or novelty concern; the prior Eulerian studies [18,46] are appropriately cited. If the authors can validate the flux sign against an Eulerian budget, acceptance would be straightforward."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The new thing here is real: a Lagrangian measurement of the kinetic energy flux across scales in a Rayleigh–Bénard cell, using a large 3D-PTV dataset. The authors show a clean sign flip in mean ⟨k̇⟩_r from negative at small scales to positive at r†≈0.32L, and they find that the flow topology associated with downscale and upscale events is not time-reversed. That is a genuinely interesting experimental observation, and the bin statistics (tens of millions of samples) give the results a solid statistical base.\n\nThe derivation from Navier–Stokes to Eq. (5) is standard, and the paper is honest about the approximations, including the neglect of strain terms for r≫η. The structure-function analysis is careful, and the authors correctly note that their measured ζ=1.3 is slightly above BO59 and attribute it to LSC shear.\n\nNow the soft spots. The most concrete one is methodological: the paper never states how k̇ is estimated from discrete 3D-PTV trajectories. No mention of differentiation schemes, filtering, or trajectory smoothing. The sign of the mean can be sensitive to the estimator, especially for a quantity dominated by intermittent events, and this is exactly the load-bearing measurement. The absence of error bars on the moments in Fig. 2 is a related problem; without them, the claim that r†=0.32±0.01L is hard to verify.\n\nThe second soft spot is more conceptual and, in my view, the one that needs the most attention. The identification of ⟨k̇⟩_r as an interscale energy flux assumes statistical homogeneity and no mean-flow contribution. This experiment has a strong persistent large-scale circulation, and at separations 0.3–0.5L the relative velocity is likely dominated by LSC shear. The authors themselves invoke LSC shear to explain the super-BO59 slope. It is not implausible, then, that the positive mean ⟨k̇⟩_r at large r is a mean-shear production term, not an upscale turbulent cascade. The stress-test note is not a straw man; it lands. The paper needs a check: subtract the local mean or large-scale field and see if the sign flip survives, or validate the Lagrangian flux against an Eulerian budget. Also, because the measurement volume excludes the near-wall regions, the boundary term ∇⟨T⟩=0 requires more discussion than a one-line Stokes argument.\n\nThe third point is minor: associating r† with the Bolgiano scale is inferred from the structure-function slope, not directly tested. That is fine as a plausibility argument, but the text should be clear it is an association, not a measured identity.\n\nWho is this for? People working on convection and energy cascade phenomenology. It is a short paper with a strong central observation and two fixable weaknesses. I would send it to peer review, with the expectation of a major revision requiring methodology details and a mean-subtraction or budget check. If those checks hold, it will be a useful paper; if they fail, the split-cascade conclusion will need to be softened.","headline":"First experimental Lagrangian flux measurement in a RB cell, with a plausible split-cascade signal, but the missing kdot methodology and the strong LSC shear make the sign-flip claim less than fully verified.","tokens_in":9829,"tokens_out":2557,"would_cite":true,"duration_ms":29874,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A split kinetic-energy cascade exists in Rayleigh-Bénard convection","keywords":["kinetic energy cascade","Rayleigh-Bénard convection","Lagrangian turbulence","particle tracking velocimetry","upscale energy transfer","downscale energy transfer","Bolgiano scale"],"falsifier":"Regenerate the same $\\langle\\dot{k}\\rangle_r$ statistics from a direct numerical simulation of the same Rayleigh-Bénard setup, sampling synthetic particles with the same density, noise, and interrogation volume; if a well-resolved material-derivative estimate does not change sign from negative to positive near $r\\approx0.3L$, the crossover would be an artefact of the trajectory estimator rather than a property of the flow.","tokens_in":8851,"feed_emoji":"🌊","tokens_out":5608,"duration_ms":51790,"temperature":0.7,"pith_summary":"This paper tries to establish that the kinetic energy cascade in turbulent Rayleigh-Bénard convection is split by scale, not uniform: on average kinetic energy moves toward smaller scales for small particle separations and toward larger scales for separations above about $r^\\dagger = 0.32L$, with a gradual mixed regime in between. The evidence comes from Lagrangian particle tracking in a cubic convection cell at $\\mathrm{Ra}=2.5\\times10^9$, following the rate of change of the kinetic energy of relative motion of tracer pairs. If the claim is correct, it tells where the energy injected by buoyancy goes and shows that the mechanisms driving downscale and upscale transfer are different physical processes, not time-reversed versions of each other. A sympathetic reader would care because it resolves a basic question about energy pathways in a canonical turbulent flow.","feed_headline":"Convection turbulence splits its energy cascade by scale","feed_subtitle":"Tracer tracking shows energy flows downscale at small scales and upscale at large scales in Rayleigh-Bénard convection.","key_machinery":"The central object is the Lagrangian flux $\\langle\\dot{k}\\rangle_r$, the average rate of change of $k = \\frac{1}{2}\\delta_r v^2$, the kinetic energy of the relative velocity of two tracer particles separated by distance $r$. The paper derives $\\langle\\dot{k}\\rangle_r = 2\\langle v\\,\\nabla p^*\\rangle_r - 2\\langle v\\, f_b^*\\rangle_r$ for scales above the dissipation range, so the flux is set by spatial correlations between velocity and pressure and between velocity and buoyancy across distance $r$. This quantity is measured from 3D particle tracking data and combined with the conditional mean of $\\dot{k}$ given the particles' separation velocity $\\delta v_r$ to identify which flow topology carries the transfer.","core_discovery":"Using the mean rate of change of the relative kinetic energy $\\langle\\dot{k}\\rangle_r$ of tracer pairs as the Lagrangian measure of kinetic energy flux across scale $r$, the paper finds a sign flip with scale. For separations $r<r_*\\approx0.06L$ the mean flux is negative and decreasing with $r$, a downscale cascade reminiscent of Kolmogorov turbulence. For $r>r^\\dagger=0.32L$ the mean flux is positive, an upscale cascade driven by thermal plumes; the crossover sits near the Bolgiano scale $L_B\\approx0.3L$, and between $r_*$ and $r^\\dagger$ the two behaviours coexist in a mixed regime in which extreme events carry the downscale flux and weaker events carry the upscale flux. The paper also claims that the flow topology is reversed between regimes: at small scales the downscale flux is associated with converging trajectories (bi-axial strain), while at large scales the upscale flux is associated with converging trajectories and the residual downscale flux with separating trajectories. This reversal is the paper's main qualitative discovery.","pith_inferences":["One testable extension is to repeat the same analysis at higher Rayleigh numbers: the prediction would be that $r^\\dagger$ stays tied to the Bolgiano scale $L_B$ rather than to a fixed fraction of $L$, so the crossover should move with $L_B$.","The conditional-topology asymmetry suggests a subgrid-modelling route: models that only parameterise downscale transfer would miss the upscale branch, so a two-way energy transfer closure may be needed for convection at these parameters.","Because the paper's flux derivation is Lagrangian, the same pair statistics could be extracted from direct numerical simulation datasets with known ground truth, which would isolate how much of the measured crossover is physical versus an artefact of discrete trajectory differentiation."],"forward_implications":["The inertial range in this flow does not have a constant flux: $\\langle\\dot{k}\\rangle_r$ keeps changing across the K41 range, so the classical picture of scale-invariant transfer does not hold there.","The scale $r^\\dagger\\approx0.32L$ is a genuine crossover of the mean flux; above it, kinetic energy is on average collected into larger scales, which sustains the large-scale circulation from below rather than only from boundary forcing.","The downscale and upscale transfer mechanisms are not time-reversed images, implying that irreversibility is visible already in the two-point Lagrangian statistics.","The gradual mixed regime between $r_*$ and $L_B$ means that local events of opposite cascade direction coexist over a broad range, not as a sharp switch."],"supporting_citations":[{"why":"Supplies the definition of $\\langle\\dot{k}\\rangle_r$ as the kinetic energy flux from relative turbulent diffusion.","marker":"[32]"},{"why":"Provides the Lagrangian view of energy transfer that the derivation follows.","marker":"[28]"},{"why":"Identifies the mean rate of change of relative kinetic energy as the flux and gives the homogeneous isotropic turbulence expectation.","marker":"[33]"},{"why":"Supports the interpretation of extreme $\\dot{k}$ events and the third-moment statistics.","marker":"[34]"},{"why":"Prior Eulerian evidence of a split cascade that this work complements with Lagrangian measurements.","marker":"[18]"},{"why":"Provides the Bolgiano-scale framework in confined convection used to interpret the upscale regime.","marker":"[46]"},{"why":"Gives the mean-field scaling regimes for Rayleigh-Bénard structure functions used to locate the inertial range and $L_B$.","marker":"[15]"},{"why":"Source of the experimental dataset the analysis is based on.","marker":"[39]"},{"why":"Particle tracking software used to obtain the trajectories.","marker":"[37]"},{"why":"Supports the association of downscale flux with straining regions in the flow-topology interpretation.","marker":"[23]"}],"fun_headline_variants":["Convection cascade splits by scale, reverses direction","Turbulent convection shows dual energy cascade","Energy flows both ways in convection turbulence","Split cascade found in Rayleigh-Bénard convection","Lagrangian tracking reveals split energy cascade"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that the value of $\\dot{k}$ computed from discrete particle trajectories is equal to the true material derivative of the relative kinetic energy of fluid elements; since the paper does not state how the trajectories were differentiated or filtered, the sign of the mean flux at small separations could depend on that unstated estimator.","fun_headline_variants_meta":{"raw":{"variants":["Convection cascade splits by scale, reverses direction","Turbulent convection shows dual energy cascade","Energy flows both ways in convection turbulence","Split cascade found in Rayleigh-Bénard convection","Lagrangian tracking reveals split energy cascade"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000137,"raw_usage":{"total_tokens":1104,"prompt_tokens":850,"completion_tokens":254,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":466,"completion_tokens_details":{"reasoning_tokens":188}},"tokens_in":466,"tokens_out":254,"duration_ms":2810,"temperature":1.0,"reasoning_tokens":188,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T04:11:57.547798+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Regenerate the same $\\langle\\dot{k}\\rangle_r$ statistics from a direct numerical simulation of the same Rayleigh-Bénard setup, sampling synthetic particles with the same density, noise, and interrogation volume; if a well-resolved material-derivative estimate does not change sign from negative to positive near $r\\approx0.3L$, the crossover would be an artefact of the trajectory estimator rather than a property of the flow.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the definition of $\\langle\\dot{k}\\rangle_r$ as the kinetic energy flux from relative turbulent diffusion."},{"cited_title":"The La- grangian view of energy transfer in turbulent flow,","cited_arxiv_id":null,"evidence_quote":"Provides the Lagrangian view of energy transfer that the derivation follows."},{"cited_title":"Parti- cles and fields in fluid turbulence,","cited_arxiv_id":null,"evidence_quote":"Identifies the mean rate of change of relative kinetic energy as the flux and gives the homogeneous isotropic turbulence expectation."},{"cited_title":"Flight– crash events in turbulence,","cited_arxiv_id":null,"evidence_quote":"Supports the interpretation of extreme $\\dot{k}$ events and the third-moment statistics."},{"cited_title":"Physical and scale-by-scale analysis of Rayleigh-B´ enard convection,","cited_arxiv_id":null,"evidence_quote":"Prior Eulerian evidence of a split cascade that this work complements with Lagrangian measurements."},{"cited_title":"Bolgiano scale in confined Rayleigh–Taylor turbulence,","cited_arxiv_id":null,"evidence_quote":"Provides the Bolgiano-scale framework in confined convection used to interpret the upscale regime."},{"cited_title":"Small-scale properties of tur- bulent Rayleigh-B´ enard convection,","cited_arxiv_id":null,"evidence_quote":"Gives the mean-field scaling regimes for Rayleigh-Bénard structure functions used to locate the inertial range and $L_B$."},{"cited_title":"Large-scale reorientation in cubic Rayleigh–B´ enard convection measured with parti- cle tracking velocimetry,","cited_arxiv_id":null,"evidence_quote":"Source of the experimental dataset the analysis is based on."},{"cited_title":"proptv: A probability-based particle tracking ve- locimetry framework,","cited_arxiv_id":null,"evidence_quote":"Particle tracking software used to obtain the trajectories."},{"cited_title":"Local energy flux and subgrid-scale statistics in three-dimensional turbulence,","cited_arxiv_id":null,"evidence_quote":"Supports the association of downscale flux with straining regions in the flow-topology interpretation."}],"review_version":1}